The link between response time and preference, variance and processing heterogeneity in stated choice experiments

In this article we utilize the time respondents require to answer a self-administered online stated preference survey. While the effects of response time have been previously explored, this article proposes a different approach that explicitly recognizes the highly equivocal relationship between response time and respondents' choices. In particular, we attempt to disentangle preference, variance and processing heterogeneity and explore whether response time helps to explain these three types of heterogeneity. For this, we divide the data (ordered by response time) into approximately equal-sized subsets, and then derive different class membership probabilities for each subset. We estimate a large number of candidate models and subsequently conduct a frequentist-based model averaging approach using information criteria to derive weights of evidence for each model. Our findings show a clear link between response time and utility coefficients, error variance and processing strategies. Our results thus emphasize the importance of considering response time when modeling stated choice data.